摘要
智能交通系统的建设已成为城市交通发展面临的主要问题,其中公交车到站时间预测是智能交通系统的重要组成部分。公交车到站时间数据是具有长期和短期特性的时间序列数据,而且公交车易受到外来因素的影响,因此公交车到站时间也是动态变化的。基于上述问题,提出基于LSTM和Kalman滤波的公交车到站时间预测模型,其中LSTM模型用来预测公交车到站的基础时间序列,Kalman滤波模型用于对基础时间数据序列进行动态调整,最终将调整后的预测值的正确率、均方差、平均绝对偏差分别与LSTM、SVM、SVM+Kalman模型预测结果进行对比,证明LSTM+Kalman模型预测值的正确率,均方差和平均绝对偏差均优于对比模型。
The construction of intelligent transportation system has become the main problem facing the development of urban transportation.The prediction of bus arrival time is an important part of intelligent transportation system.Bus arrival time data is time-series data with long-term and short-term characteristics,and the bus is susceptible to external factors,so the bus arrival time is also dynamic.Based on the above problems,a bus arrival time prediction model based on LSTM and Kalman filtering was proposed,in which LSTM model was used to predict the basic time series of bus arrival and arrival,and Kalman filtering model was used to dynamically adjust the basic time series.Finally,the adjusted prediction accuracy,mean square deviation and mean absolute deviation were respectively compared with those predicted by LSTM,SVM and SVM+Kalman models.It was proved that the accuracy,mean square deviation and average absolute deviation of LSTM+Kalman model prediction were better than the comparative model.
作者
范光鹏
孙仁诚
邵峰晶
Fan Guangpeng;Sun Rencheng ;Shao Fengjing(College of Computer Science and Technology,Qingdao University,Qingdao 266071,Shandong,China)
出处
《计算机应用与软件》
北大核心
2018年第4期91-96,共6页
Computer Applications and Software
基金
国家自然科学基金面上项目(41476101)